From: Gerry Sheng <*shengyc*>

Date: Sat, 20 Feb 2016 22:25:43 +0000

Hi Mark,

My first suggestion is you can start from simpler mixture model (e.g. 2

distributions) and only focus on those have AEs. 68% patients without AE is

a big disturbance to intercept. Only negative infinity of intercept in

logistic model can give a probability=0. Secondly, you can try to use SAE=

M

method with mu reference. Based on my own experience, FOCE method is not as

powerful as SAEM in likelihood estimations. Good luck.

BW,

Yucheng Sheng

UCL School of Pharmacy

29-39 Brunswick Square

London WC1N 1AX

Email. shengyc

On 19 February 2016 at 22:30, Mark Sale <msale

*> Has anyone every tried to use a mixture model with logistic regression? I
*

*> have data on a AE in several hundred patients, measured multiple times
*

*> (10-20 times per patient). Examining the data it is clear that,
*

*> independent of drug concentration, there is very wide distribution of thi=
*

s

*> AE, 68% of the patients never have the AE, 25% have it about 20% of the
*

*> time and the rest have it pretty much continuously, regardless of
*

*> drug concentration. (in ordinary logistic regression, just glm in R, the=
*

re

*> is also a nice concentration effect on the AE in addition). Running the
*

*> usual logistic model, not surprisingly, I get a really big ETA on the
*

*> intercept, with 68% of the people having ETA small negative, 25% ETA ~ 1
*

*> and 7% ETA ~ 10. No covariates seem particularly predictive of the post h=
*

oc

*> ETA. I thought I could use a mixture model, with 3 modes, but it refused
*

*> to do that, giving me essentially 0% in the 2nd and 3rd distribution, sti=
*

ll

*> with the really large OMEGA for the intercept. Even when I FIX the OMEGA
*

*> to a reasonable number, I still get essentially no one in the 2nd and 3rd
*

*> distribution. I tried fixing the fraction in the 2nd and 3rd distributio=
*

n

*> (and OMEGA), and it still gave me a very small difference in the intercep=
*

t

*> for the 2nd and 3rd populations.
*

*>
*

*> Is there an issue with using mixture models with logistic regression? I'm
*

*> just using FOCE, Laplacian, without interaction, and LIKE.
*

*>
*

*>
*

*>
*

*>
*

*> Any ideas?
*

*>
*

*>
*

*> Mark
*

*>
*

*>
*

*>
*

*> Mark Sale M.D.
*

*>
*

*> Vice President, Modeling and Simulation
*

*>
*

*> Nuventra, Inc. ™
*

*>
*

*> 2525 Meridian Parkway, Suite 280
*

*>
*

*> Research Triangle Park, NC 27713
*

*>
*

*> Office (919)-973-0383
*

*>
*

*> msale *

*>
*

*> www.nuventra.com
*

*>
*

*>
*

*>
*

*>
*

Received on Sat Feb 20 2016 - 17:25:43 EST

Date: Sat, 20 Feb 2016 22:25:43 +0000

Hi Mark,

My first suggestion is you can start from simpler mixture model (e.g. 2

distributions) and only focus on those have AEs. 68% patients without AE is

a big disturbance to intercept. Only negative infinity of intercept in

logistic model can give a probability=0. Secondly, you can try to use SAE=

M

method with mu reference. Based on my own experience, FOCE method is not as

powerful as SAEM in likelihood estimations. Good luck.

BW,

Yucheng Sheng

UCL School of Pharmacy

29-39 Brunswick Square

London WC1N 1AX

Email. shengyc

On 19 February 2016 at 22:30, Mark Sale <msale

s

re

oc

ll

n

t

Received on Sat Feb 20 2016 - 17:25:43 EST